marek/MtoRagSystem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

optimize py autoload

Browse Source
This commit is contained in:
team2
2026-02-26 17:43:22 +01:00
parent deba7cd06f
commit 5b4e0ad4c3
3 changed files with 131 additions and 1879 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
src/Knowledge/Retrieval/NdjsonHybridRetriever.php
View File

@@ -6,6 +6,7 @@ namespace App\Knowledge\Retrieval;
use App\Entity\ModelGenerationConfig;
use App\Knowledge\ChunkManager;
use App\Knowledge\QueryCleaner;
use App\Repository\ModelGenerationConfigRepository;
use App\Tag\TagRoutingService;
use App\Vector\VectorSearchClient;
@@ -24,6 +25,7 @@ final class NdjsonHybridRetriever implements RetrieverInterface
private readonly VectorSearchClient $vectorClient,
private readonly TagRoutingService $tagRouting,
private readonly ModelGenerationConfigRepository $configRepository,
private readonly QueryCleaner $queryCleaner,
) {}
public function retrieve(string $prompt): array
@@ -47,12 +49,21 @@ final class NdjsonHybridRetriever implements RetrieverInterface
$limit = max(1, min($config->getRetrievalMaxChunks(), self::HARD_MAX_CHUNKS));
$vectorTopKBase = max(1, min($config->getRetrievalVectorTopK(), self::HARD_MAX_VECTORK));
// Wichtig: List-Detection bleibt auf Originalprompt (sonst entfernst du "zeige/liste" etc.)
$isListQuery = $this->isListQuery($prompt);
// -------------------------------------------------
// 1) Tag Routing
// CLEAN QUERY (nur für Retrieval: Tags + Vector)
// -------------------------------------------------
$candidateDocIds = $this->tagRouting->route($prompt);
$cleanQuery = $this->queryCleaner->clean($prompt);
if ($cleanQuery === '') {
$cleanQuery = $prompt;
}
// -------------------------------------------------
// 1) Tag Routing (bereinigte Query)
// -------------------------------------------------
$candidateDocIds = $this->tagRouting->route($cleanQuery);
$candidateSet = null;
if (is_array($candidateDocIds) && $candidateDocIds !== []) {
@@ -76,22 +87,22 @@ final class NdjsonHybridRetriever implements RetrieverInterface
}
// -------------------------------------------------
// 3) Vector Search (Scoped wenn möglich)
// 3) Vector Search (bereinigte Query; scoped wenn möglich)
// -------------------------------------------------
if ($candidateSet !== null) {
$hits = $this->vectorClient->searchScoped(
$prompt,
$cleanQuery,
$topK,
array_keys($candidateSet)
);
// Wenn scoped nichts liefert → global fallback
if ($hits === []) {
$hits = $this->vectorClient->search($prompt, $vectorTopKBase);
$hits = $this->vectorClient->search($cleanQuery, $vectorTopKBase);
}
} else {
$hits = $this->vectorClient->search($prompt, $topK);
$hits = $this->vectorClient->search($cleanQuery, $topK);
}
if ($hits === []) {
@@ -294,7 +305,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
}
$chunk = trim($rows[$id]['text']);
$key = mb_strtolower((string)preg_replace('/\s+/u', ' ', $chunk));
if (isset($seen[$key])) {